Parse lines at a time

examples/test.modal

@@ -1,4 +1,5 @@
 <> (?x dup) (?x ?x)
 <> (?x ?y swap) (?y ?x)
 
-(A B swap) dup
+((A B swap) dup)
+

src/modal.c

@@ -112,62 +112,59 @@ print_rules(void)
 }
 
 static char *
-parse_rulefrag(FILE *f)
+parse_rulefrag(char *line)
 {
-	int depth = 0, trim = 0;
-	char c, *origin = dict_;
-	while(f && fread(&c, 1, 1, f) && c && c != 0xa) {
-		if(c == ' ' && !trim) continue;
-		trim = 1;
-		if(c == '(') {
-			depth++;
-			if(depth == 1) continue;
-		}
-		if(c == ')') {
-			--depth;
-			if(depth == 0) continue;
-		}
-		if(c == ' ' && !depth) break;
+	int depth = 0;
+	char c, *s = line, *res = dict_;
+	if(s[0] == '(') {
+		while((c = *s++)) {
+			if(c == '(') depth++;
 			*dict_++ = c;
-	}
+			if(c == ')') --depth;
+			if(!depth) {
 				*dict_++ = 0;
-	return origin;
+				return res;
+			}
+		}
+	}
+	while(!spacer(s[0]) && (*dict_++ = *s++))
+		;
+	*dict_++ = 0;
+	return res;
 }
 
-static void
-tokenize(char *t, FILE *f)
+static int
+parse_line(char *line)
 {
-	char c;
-	if(!t[0]) return;
-	if(t[0] == '<' && t[1] == '>') {
+	if(line[0] == 0) return 1;
+	if(line[0] == '<' && line[1] == '>') {
 		Rule *r = &rules[rules_len++];
-		r->a = parse_rulefrag(f), r->b = parse_rulefrag(f);
-		return;
+		line += 3;
+		r->a = parse_rulefrag(line);
+		line = walk(line) + 1;
+		r->b = parse_rulefrag(line);
+		return 1;
 	}
-	while((c = *t++))
-		*prog_++ = c;
-	*prog_++ = ' ';
-	while(f && fread(&c, 1, 1, f) && c)
-		*prog_++ = c == 0xa ? ' ' : c;
+	while((*prog_++ = *line++))
+		;
+	*prog_++ = 0xa;
+	return 1;
 }
 
 static int
 parse(char *path)
 {
 	FILE *f;
-	char c, token[0x40], *token_;
+	char c, line[0x400], *line_ = line;
 	if(!(f = fopen(path, "r")))
 		return !printf("Invalid file: %s\n", path);
-	token_ = token;
 	while(f && fread(&c, 1, 1, f)) {
-		if(c < 0x21)
-			*token_++ = 0x00, tokenize(token, f), token_ = token;
-		else if(token_ - token < 0x3f)
-			*token_++ = c;
-		else
-			return printf("Token too long: %s\n", token);
+		if(c == 0xa)
+			*line_++ = 0x00, parse_line(line), line_ = line;
+		else if(line_ - line < 0x400)
+			*line_++ = c;
 	}
-	*token_++ = 0x00, tokenize(token, f), token_ = token;
+	*line_++ = 0x00, parse_line(line), line_ = line;
 	fclose(f);
 	return 1;
 }